WarehouseOps editorial · Updated 20 September 2026
Define the population
Use the same orders and promised delivery dates for on-time, in-full and both. Agree whether partial deliveries, customer holds and cancellations belong in the denominator. Do not combine dispatched orders for one measure with delivered orders for another.
Worked example
Of 100 orders, 80 are on time and 90 are in full. If 75 satisfy both, OTIF is 75%. If the overlap was not measured, it could be between 70 and 80 orders. Multiplying 80% by 90% gives 72%, which is an assumption about overlap, not a measurement.
When zero is real
Of 100 orders, 40 may be on time and 30 in full with no order satisfying both. Measured OTIF is then 0%. Leaving the overlap blank means unknown, whose feasible range is 0–30% in this example.
Investigate before assigning blame
Take a sample of failed orders and code the actual reason: supply, allocation, picking, dispatch, carrier, appointment or promise-date error. The two totals alone do not establish who caused failure.